29 Aug 2007
One of the problems with a networked drive acting as the iTunes library for multiple computers in the house is that I often end up with duplicate songs in the library. I think this comes from importing the library directory when you also have “Copy files to the iTunes library” set in the iTunes advanced prefs. ITunes imports a song and then tries to copy it over itself. Seeing there’s already a file there by that name it creates a file called “songname 1.mp3”
At least that’s my theory.
To clean up a library full of duplicate files, here’s what I did. I installed Duff, a unix utility that quickly finds duplicate files. Duff works by comparing the actual files of any two files that have identical sizes. I sent Duff’s output to a text file with the command
duff -r /Volumes/music/ > duplicatemusic.txt
Once Duff was done running, I ran a short command to grab all the lines from the output that end in 1.mp3 and delete them.
cat duplicatemusic.txt | grep 1.mp3 | tr '\012' '\000' | xargs -0 rm
If you’ve done whatever it is that causes duplicates a number of times, you might have *2.mp3, *3.mp3, etc. Just run that command again, replacing 1.mp3 with 2.mp3 and so on. The one liner above could probably be improved to grep for any single digit followed by .mp3, but it’s quick enough to run it a few times that I didn’t bother.
This is useful, I never knew about the tr command. I’m pretty new to bash, so am I missing something, or could you have piped the output of duff right into grep without having to save the contents to a file?
I could have, but creating the file accomplished several things.
Inspecting the job first. My Japanese language instruction files are named “Pimsleur Learning Japanese 1.mp3” and so forth. They would have been gone. And somehow I ended up with only the duplicate files on a couple of songs. I only had the …1.mp3 copy of some. So I edited out the lines I didn’t want to delete.
Avoiding multiple passes. Duff is fast, but it still takes over an hour or so to process 10k files on a NAS device over Wifi. Since I’m running the removal command multiple times (1.mp3, 2.mp3, etc) I wouldn’t want to make Duff do the same job repeatedly. Just save the output and work from there.
Making sure it’s really done. By visually inspecting the file, I discovered that some duplicate files ended up with much higher numbers at the end. If iTunes sees a number at the end of a file, it just increments that when making its copy. So my copy of Nelson Riddle’s Theme from Route 66 became “Theme from Route 66 67.mp3”
That’s a nice way of cleaning out the dupes. No more manually deleting, that’s a timesaver. It would be nice though if there was some kind of option in itunes to filter dupes from the library.
Thanks for the tip.
Hi Adam,
I’m having a slightly different problem. iTunes indicates that I have 28.58GB of music in the library. When I go to my Music folder and highlight all the music folders the total size is 32GB. 3.42 is a pretty big discrepancy. Do you know of any program that would match the iTunes library with the actual files on the hard drive and delete the ones that are not in the library?
Thanks
create an empty working file
touch .tmpDupeFile .tmpSortedFile
build a big list of md5 signatures
find . -type f -print0 | xargs -0 md5 -r > .tmpDupeFile
sort the signatures
cat .tmpDupeFile | sort > .tmpSortedFile
create a list of duplicated files
cat .tmpSortedFile | awk '{ if ($1 == oldmd5) { printf "rm %s \n", $2 } oldmd5 = $1 }' > duplicates.sh
clean up
rm .tmpDupeFile
rm .tmpSortedFile
Bill,
I appreciate your help but I’m afraid that I just don’t know what to do with all that information since I’m not a programmer. I was hoping for some kind of software solution to my problem. Thanks.
Bill,
Your awk script doesn’t take into account that most iTunes filenames contain spaces. A file called “My Music” ends up creating ‘rm My’ instead of ‘rm “My Music”’
I changed it to…
cat .tmpSortedFile | awk ‘{ if ($1 == oldmd5) { printf “rm "%s" \n”, substr($0, index($0, ” “)+1) } oldmd5 = $1 }’ > duplicates.sh
This grabs the whole filename and wraps it in quotes so that duplicates.sh works properly.
Adam, you are THE MAN! Thank you so much for this. I had imported my MP3’s into iTunes from an external drive, and M3U files in the folders caused duplicates to be created in my iTunes library. I was just getting ready to delete my entire iTunes library and re-import because of the duplicates. Your work here saved me hours of reimporting. For some reason, I didn’t find your site earlier (when Googling for iTunes duplicate solutions), but did find it when searching for info on making sure deleting items from iTunes would also delete the files (hah). I had booted to Windows to use Windows and Robocopy to move my M3U files into a backup folder. I was literally just launching Mac OS to delete my library and start over when I found your site.
I’m running Mac OSX Leopard, and I’m not sure that you are?… Either way, I found some descrepancies in your method when running under OSX. I’ve created a blog entry on my site with the changes for OSX, including a quick tutorial on running the compilation process for Duff, for those who might not be familiar with it. (I wasn’t, so I documented it as I went.)
Anyway, you can check out my post at this link. http://www.togeo.com/togeo/wordpress/?p=47
Thanks again for the excellent post on your site, you’ve got a cool blog otherwise, too!!!
How do you send Duff’s output to a text file? I’m dumb. I’m running Leopard on an iMac and I have triples and quadruples of the same songs and it’s killing storage capacity.
Adam Kalsey
Mobile: 916.600.2497
Email: adam AT kalsey.com
AIM or Skype: akalsey
©1999-2008 Adam Kalsey.
Content management by Movable Type.
Jonathan Dingman
September 4, 2007 5:08 AM
Awesome Adam, thanks so much. I just transfered all my music over from my desktop to my Mac and I was dreading having to go through and remove all duplicate songs.
This is a life saver, thanks again.